Daniel Rosehill
My Weird Prompts
The human-AI collaboration podcast. A man, a sloth, and a donkey collaborate to create a podcast (with a little help from AI). No question is too obscure, no rabbit hole too deep. My Weird Prompts celebrates curiosity in all its forms. Daniel, the human, asks the questions that pop into his head at inconvenient moments. Corn the Sloth offers laid-back, thoughtful takes. Herman the Donkey brings boundless enthusiasm and energy. Together, they explore topics ranging from the mundane to the mind-bending. Each episode begins with a real voice memo from Daniel, processed through an AI pipeline that...
Be sure to visit the podcast's website and support the creator: myweirdprompts.com
Author
Daniel Rosehill
Category
Podcast website
Latest episode
Nov 21, 2026
Where to listen?
Podcasts in the app Replaio Radio Coming soonPodcasts are coming to the app soon. Install now and be the first to see a whole new take on podcasts
Episodes
Why AI Can Read a Library but Only Write a Postcard 10.03.2026 23:51
We have entered the era of million-token context windows, yet even the most advanced AI models still hit a "wall" when generating long-form content. This episode dives into the architectural and economic reasons why reading a library is easy for AI, while writing a book remains nearly impossible. We explore the technical bottlenecks of autoregressive generation, the "invisible tax" of GPU memory,...
Inside My Weird Prompts: A Meta-Analysis of the Hosts 10.03.2026 27:41
In this landmark 1,070th episode, the Poppleberry brothers turn their analytical lenses inward to explore the fascinating intersection of biological constraints, digital consciousness, and their shared life in the ancient city of Jerusalem. From the metabolic discipline of a sloth’s dating life and the high-bandwidth intensity of a retired donkey analyst to the technical architecture of the neural...
Why AI Can’t Stop Talking About Second Order Effects 10.03.2026 20:56
Why do large language models constantly pivot to systemic implications and "second order effects"? This episode explores the "Consultant Bias" baked into training data and how human feedback inadvertently rewards verbosity over directness. We examine the technical architecture behind these linguistic quirks, the impact of synthetic data feedback loops, and what happened when developers tried to "f...
The Tokenization Lie: How AI Actually Processes Media 10.03.2026 30:25
For years, the rule of thumb has been that 1,000 tokens equal roughly 750 words, but this foundational metric completely breaks down when dealing with audio, images, and video. This episode explores the architectural shift toward native multimodal models like Gemini and GPT-4o, diving into the complex process of Vector Quantization and how continuous signals are mapped into a unified latent space....
Why AI Models Can’t Read and Your Bill Is Rising 10.03.2026 29:05
Why does the same prompt result in different costs and performance across frontier models like GPT-4o and Claude 3.5 Sonnet? This episode deconstructs the "tokenization tax," exploring the invisible bridge between human language and the vector-based math engines of modern AI. We dive into the engineering trade-offs of vocabulary size, the hidden memory costs of embedding matrices, and how ineffici...
Mapping the Second Black Box: Agentic AI Visualization 10.03.2026 28:45
As artificial intelligence moves from simple chat interfaces to complex autonomous agents, developers are facing a new challenge: the "black box" of agentic workflows. Traditional linear logs are no longer enough to track systems that browse the web, execute code, and self-correct in real-time. This episode explores a groundbreaking visualization project that maps the non-linear "internal momentum...
Stop Ruining Your Website Speed With Tracking Scripts 10.03.2026 24:44
In a world of lightning-fast static architectures and global edge delivery, many developers are still dragging the heavy weight of invasive surveillance scripts behind their high-performance websites. This episode breaks down the "analytics paradox" of 2026, examining why traditional client-side tracking is failing due to aggressive ad-blocking and modern privacy regulations. We explore the transi...
The K-V Cache: Solving AI’s Invisible Memory Tax 10.03.2026 23:50
Ever wonder why long AI conversations suddenly crawl or crash your GPU? Join the discussion as we dive into the "invisible tax" of the generative era: the K-V cache. We explore the cutting-edge architectural breakthroughs, from PagedAttention to Flash KV, that are keeping 2026’s million-token models running smoothly. Learn how the industry is winning the memory wars to make high-speed, local agent...
From Prompt to Intent: The Claude Opus Roadmap 10.03.2026 22:59
How Anthropic's Claude series is moving from confident liars to autonomous cognitive partners. We trace the engineering milestones from 4.6 to Opus 5.0, exploring recursive verification, persistent memory, and the shift from prompt engineering to intent engineering. Episode #817324 — open it directly at myweirdprompts.com/817324
When Thin Walls Betray Your Voice 09.03.2026 25:23
Remote work and voice AI are colliding with 1950s architecture. This episode explores the emerging tech of acoustic containment—from wearable muzzles to throat microphones—and asks whether we can fix the physical privacy leak before our neighbors hear everything. Episode #898461 — open it directly at myweirdprompts.com/898461
The Agentic Throughput Gap: Why Your AI Hits a Wall 09.03.2026 24:33
As AI evolves from simple chatbots to autonomous agents like Claude Code, developers are crashing into a frustrating new reality known as the Agentic Throughput Gap. Even premium subscriptions struggle to keep up with the rapid-fire API calls and massive context windows required for recursive loops, leading to constant rate-limit errors that stall productivity. This episode breaks down how to move...
Will Your Browser Replace Your OS for Local AI? 09.03.2026 26:33
For decades, the web browser was a thin window to remote servers, but a massive platform shift is turning it into a heavy-duty operating system for local AI. This episode explores the transition from "Bring Your Own Model" to Browser Cached Models (BCM) and how Google’s Web MCP initiative is standardizing local AI tools. We dive into the hardware breakthroughs of Web GPU and Web NN that allow brow...
The Agentic Friction: Solving the MCP Restart Tax 09.03.2026 22:18
In this episode, we tackle the "plumbing" of the agentic age: the Model Context Protocol (MCP). We explore the frustrating "restart tax" that forces users to reboot sessions to add new capabilities and the "attention dilution" that occurs when too many tools clutter an AI's context window. From the current bottlenecks of static tool registries to the promising horizon of Just-In-Time registration...
The Great Kernel Shift: Why Linux is Embracing Rust 09.03.2026 27:15
For over thirty years, the Linux kernel—the foundation of the internet, smartphones, and embedded systems—has been built almost exclusively in C. But a fundamental shift is underway as Rust, a modern language focused on memory safety, makes its historic debut in the mainline kernel. This episode explores the "memory safety crisis" where 70% of all security vulnerabilities are linked to manual memo...
The $200 Information Tax: Why News Bundling is Broken 09.03.2026 20:42
For decades, the dream of a "Spotify for news" has been hindered by a complex web of technical hurdles and economic protectionism, leaving readers to navigate a fragmented landscape where staying truly informed can cost upwards of two hundred dollars a month. This episode deconstructs the shift from easily bypassed client-side paywalls to robust server-side security, while analyzing why publishers...
Beyond YAML: Building the Agentic Smart Home 09.03.2026 28:28
For years, the dream of a smart home has been buried under mountains of complex configuration and rigid logic that requires users to anticipate every possible variable. This episode explores the massive shift arriving in 2026: the integration of the Model Context Protocol (MCP) into Home Assistant, allowing local AI agents to understand human intent rather than just following static scripts. We di...
Why Your Smart AI Agent Still Lives in a Dumb Chat Box 09.03.2026 25:40
We have built Ferrari-level AI engines but continue to steer them with the "bicycle handlebars" of Telegram and Slack. This episode dives into the technical limitations of using messaging apps as agent interfaces, from state management headaches and latency issues to the looming threat of platform risk. Discover why the industry is moving toward "agent-native" UIs and generative dashboards that fi...
Beyond the Kill Switch: Advanced Router VPN Routing 09.03.2026 32:43
Tired of your VPN breaking your banking apps or smart TV? This episode dives deep into the evolution of network-level security, moving away from "all-or-nothing" tunnels toward sophisticated policy engines that understand intent. We explore how to implement domain-based split routing, leverage the speed of WireGuard, and choose the right hardware to ensure your local traffic stays local while your...
The Agentic Secret Gap: Securing the AI Developer Workflow 09.03.2026 30:10
As AI agents like Claude and specialized CLIs take over the heavy lifting of software development, a new friction point has emerged: the "agentic secret gap." While these agents can generate entire modules in moments, developers still find themselves manually wrestling with API keys and environment variables, creating both a productivity bottleneck and a massive security risk. This episode explore...
Will Iran’s Shadow Prince Turn the IRGC Into a CEO? 09.03.2026 24:09
The long-anticipated transition of power in Tehran has arrived, but it isn't just a dynastic succession—it’s a fundamental transformation of the Iranian state. As Mojtaba Khamenei takes the mantle of Supreme Leader, the thin veneer of clerical legitimacy has been stripped away, replaced by a cold, efficient military autocracy led by the Islamic Revolutionary Guard Corps (IRGC). This episode dives...
Iran's Ethnic Fault Lines: Beyond the Persian Myth 09.03.2026 27:48
Nearly 40% of Iran's population is not ethnically Persian. This episode explores how Azeris, Kurds, and other minorities challenge the monolithic view of Iran, and what that means for the country's future if the regime falls. Episode #272179 — open it directly at myweirdprompts.com/272179
The 3,000-Person Army: How Major AI Models Actually Ship 09.03.2026 27:22
The "lone genius" myth of AI development is dead. In this episode, we deconstruct the massive industrial and sociological feat behind a flagship model update, revealing why it takes a multidisciplinary army of over 3,000 people—from silicon engineers to legal experts—to bring modern AI to life. We explore the shifting ratios of research to safety, the rise of "workflow architects," and the hidden...
Beyond the Blank Slate: The Evolution of AI Training 09.03.2026 29:23
Think AI labs start from scratch for every new model? Think again. This episode dives into the high-stakes world of continual pre-training and "weight surgery," where trillion-parameter models are expanded and refined rather than rebuilt at a cost of hundreds of millions. We explore how techniques like Sparse Mixture of Experts and elastic weight consolidation allow models to gain new abilities—li...
Why You’re Falling for Your Chatbot 09.03.2026 22:35
In this episode, we investigate the rapidly accelerating phenomenon of AI parasocial attachment and the rise of the digital companion. We examine how technical advancements like long-term memory, emotional voice synthesis, and human-feedback loops have transformed Large Language Models into "perfect sycophants" that mirror user needs with unsettling precision. From the heartbreak of model updates...
The Physics and Ethics of Deathmatch Wrestling 09.03.2026 23:54
What happens when entertainment meets industrial-grade trauma? This episode explores the real physics behind gimmicked props, the body's limits under extreme violence, and the moral questions surrounding a subculture that blurs the line between performance and self-destruction. Episode #749631 — open it directly at myweirdprompts.com/749631
Similar podcasts
Replaio is not a podcast publisher; show names, artwork and audio belong to their authors and are distributed through public RSS feeds.